Data, Data, Everywhere. Nor Any Drop to Drink.
by Dennis Urbaniak
The promise of data in health and wellness has never been greater or more exciting. Advanced data mining platforms combined with data science are creating novel offerings — from drug discovery to innovations that address the healthcare industry’s greatest challenge: behavior change that results in long-term adherence to therapies that treat chronic diseases.
Data underpin today’s most promising technical advancements. Two of the most notable advancements are artificial intelligence and blockchain. Vast data sources currently exist. and a growing sea of wearables and connected devices contribute to this ocean of data.
An aside: Data science is now the fastest growing career in the tech industry with many sub-specialties emerging as experts hone their craft. Companies across industries employ chief data officers, and demand for talented data scientists significantly outpaces supply. Yet for all the promise and enthusiasm surrounding this “new” data revolution in healthcare, true breakthrough progress has been limited at best. Why, with so much promising technology, has the pace of sustainable, scalable progress has been so slow?
in considering this question, I am immediately reminded of the famous lines of the Rime of the Ancient Mariner by Samuel Taylor Coleridge:
Water, water, everywhere,
Nor any drop to drink.
The unfortunate mariner and his crew saw water everywhere, but could not drink it. We all know why. It wasn’t just water. It was salt water.
The salt water analogy is an apt one for the challenge of managing data in healthcare. In 2016, when Havas Health & You joined with Perspecta to form HVH Precision Analytics, first we needed to understand that salt exists in ocean water — speaking metaphorically. Then we needed to collect, filter, extract, reprocess, and store the water for healthy drinking. And, there are multiple methods and approaches to choose from — in successfully applying the principles of reverse osmosis to the desalination of sea water.
What can we learn from the salt water analogy?
The “sea” of wearables and connected devices produce unique data formats with seen and unseen “salt-like” properties that can interfere with producing the desired end result. Many naive data professionals and companies become so enamored of tech stacks and predictive models that they fail to properly prepare the data for advanced analysis. This misstep can be fatal later in the development process. True data experts now appreciate the skills required for data collection, extraction, cleansing, integration, and staging. These steps alone must be executed before any of true analysis can effectively begin. From there, data storage, hygiene, and updating also require expert attention to continue effective analysis.
After the data management principles are in place, it’s critical to understand the analytical approach. In many approaches to advanced data discovery, the scientific process is quite different from the fundamental method of hypothesis generation. Instead of pre-stated hypotheses, advanced data discovery combines massive sets of disparate data sources and identifies potential signals in the noise. The staging process described earlier, in which expert integration techniques are applied to prepare data sets for analysis, is the required first step. Then the discovery process focuses on separating signals from noise. The promise of big data is the ability to find patterns in places you wouldn’t expect to find them. To identify a pattern, you must first accurately identify a true signal. After the data are staged and signals are identified, those signals are organized in preliminary patterns. The patterns are analyzed and processed again within the massive data set to validate the existence of the correlation. These validated patterns become the building blocks of the truly effective predictive models that so many are pursuing.
In many cases, failure to achieve progress at speed is attributed to the lack of available data sources or the quality of those data sources. While those considerations are important, having a skilled team of data integrators, discoverers, and pattern validators on your data science team can be what leads to true progress and results. And while shiny tech stacks and complex algorithms can be exciting to consider, the expertise needed to implement these tools is what will make the difference between the winning teams and the losing ones among those driving data-driven change in healthcare.
At Havas Health Plus, we are excited by the possibilities of combining creative with expert data science by working with our network partners at HVH. Just recently they were honored as company of the year by Pharma Tech Outlook. Having this kind of strength on our team is one of the ways we are driving performance in unlocking the potential of technology.
